Virtual user approach for group recommender systems using precedence relations Venkateswara Rao Kagita, Arun K. Pujari, Vineet Padmanabhan ⇑ Artificial Intelligence Lab, School of Computer & Information Sciences, University of Hyderabad, Hyderabad 500046, Andhra Pradesh, India article info Article history: Received 13 March 2014 Received in revised form 10 July 2014 Accepted 26 August 2014 Available online 28 September 2014 Keywords: Recommender system Virtual user Precedence relation abstract In this paper, we propose a novel virtual user strategy using precedence relations and develop a new scheme for group recommender systems. User profiles are provided in terms of the precedence relations of items as used by group members. A virtual user for a group is constructed by taking transitive precedence of items of all members into consid- eration. The profile of the virtual user represents the combined profile of the group. There has not been any earlier attempt to define virtual user profile using precedence relations. We show that the proposed framework exhibits many interesting properties. Earlier approaches construct virtual user profile by considering the set of common items used by all members of the group. In the present work, we propose a method of computing weightage for each item, not necessarily common to all members, using transitive prece- dence. We also introduce a new measure called monotonicity to measure the perfor- mance of any recommender system. In a top-k recommendation, monotonicity tries to measure the number of items continued to be recommended when a technique is utilized incrementally. We experimented extensively for different combinations of parameter set- tings and for different group sizes on MovieLens data. We show that our framework has better performance in terms of precision and recall when compared with other meth- ods. We show that our recommendation framework exhibits robust monotonicity. Ó 2014 Elsevier Inc. All rights reserved. 1. Introduction Recommender systems have become valuable resources for users seeking advice on products, movies, web pages, etc. and are gaining widespread acceptance in web services applications such as Amazon [14], Netflix (www.netflix.com) and YouTube [5]. Recent research also shows how recommender systems are gaining importance in a University environment [25,28,29,33]. Based on what kind of recommendation technique is used, recommender systems, are usually classified into three categories [1], Collaborative Filtering (CF), Content based (CB) and Hybrid. Several recommender systems have been pro- posed in the Literature that makes use of the above strategies as well as other recommendation techniques like demographic- based [24], association rule-based [9], utility-based and knowledge-based. Of the different recommendation strategies as out- lined above two of them stand out: (1) Content-based and (2) Collaborative Filtering. The content-based filtering approach [18,24,26,27] creates a profile for each user/product to characterize its nature. Content-based strategies require gathering external information that might not be available or easy to collect. Collaborative filtering [4,10,22,31,32,35] relies only on past user behavior without requiring the creation of explicit profiles. A major appeal of collaborative filtering is that it is domain http://dx.doi.org/10.1016/j.ins.2014.08.072 0020-0255/Ó 2014 Elsevier Inc. All rights reserved. ⇑ Corresponding author. E-mail addresses: venkateswar.rao.kagita@gmail.com (V.R. Kagita), akpcs@uohyd.ernet.in (A.K. Pujari), vineetcs@uohyd.ernet.in (V. Padmanabhan). Information Sciences 294 (2015) 15–30 Contents lists available at ScienceDirect Information Sciences journal homepage: www.elsevier.com/locate/ins